The tank container was built in June 1993 to US Department of Transport (USDOT) specification IM101. Although the International Maritime Dangerous Goods (IMDG) Regulations applied at the time of the occurrence, the tank had been certificated to meet 49CFR, and, therefore, should also have met those requirements. The IMDG vertical loading requirement is 2W (where W represents the weight of the tank, its fittings and its contents) as compared with the 49CFR which states that the calculated stress of the tank support structure must not exceed 80percent of its tensile strength when the tank is subjected to a vertical inertial load factor of 3W. If the tank had been loaded to its maximum permissible weight of 29,030kg, excluding the weight of the frame, a vertical inertial load factor of between 3.25 and 3.5W would have been required to cause this failure. The actual tank weighed somewhat less at 28,246kg and would have required a slightly higher vertical inertial load factor of 3.3 to 3.6W to fail. However, because of the location of the container in the forward hold, the rough seas experienced during the storm and the speed of the ship, the container would have been subjected to multiple high impact loads because of the ship's pitching and pounding. Further, given that containers are free to move within their guides, and that only one container was stacked on top of the failed container, it is possible that the failed container was able to move vertically from its footings while the vessel was pitching and pounding, causing significant vertical impact loads. The failed container was filled to 77percent of its capacity. As per the IM101 requirements, tanks containing between 20 and 80percent of their capacity should not be offered for transport by ship, and, therefore, the failed container was marginally underfilled. This could have resulted in liquid surge, shifting the centre of gravity to one end. This effect, combined with vertical acceleration, can significantly increase the vertical loads at one end. The IMDG Code has a similar requirement, although the exact ullage allowance must be determined using a formula which takes into account the temperatures and expansion characteristics of the contents. The damage pattern on the container was consistent with the damage having occurred as a result of operational loads during a storm at sea. However, the possibility exists that the longitudinal I-beams may have been slightly buckled or otherwise damaged prior to loading. If a buckle was present on the longitudinal I-beam prior to loading aboard the ship, the load required to cause the type of failure observed would have been much less than on an undamaged frame. To reduce the damage resulting from handling in the railway transportation system, and to facilitate observation of such damage, the container manufacturer has instituted a 36-ton upgrade of the unit by welding diagonal braces in two additional holes in the web, stiffening the latter and increasing its buckling strength. To date, more than 46percent of the 1,448 tank containers produced in the 30-ton series have been upgraded to the 36-ton level. At this time, there has been no reported failure of the new 36,000kg-rated tank container frame. The chemical that spilled into the No.1 hold of the MORU.K. is listed in the IMDG Code published by the International Maritime Organization (IMO). Its properties are given under numerical index UNNo.1760. Its safety data sheet indicates that it is a Class8 product, which means it is corrosive. This solution is composed of sodium borohydride, sodium hydroxide and water, and its properties are to react chemically when exposed to metals to release hydrogen.Analysis The tank container was built in June 1993 to US Department of Transport (USDOT) specification IM101. Although the International Maritime Dangerous Goods (IMDG) Regulations applied at the time of the occurrence, the tank had been certificated to meet 49CFR, and, therefore, should also have met those requirements. The IMDG vertical loading requirement is 2W (where W represents the weight of the tank, its fittings and its contents) as compared with the 49CFR which states that the calculated stress of the tank support structure must not exceed 80percent of its tensile strength when the tank is subjected to a vertical inertial load factor of 3W. If the tank had been loaded to its maximum permissible weight of 29,030kg, excluding the weight of the frame, a vertical inertial load factor of between 3.25 and 3.5W would have been required to cause this failure. The actual tank weighed somewhat less at 28,246kg and would have required a slightly higher vertical inertial load factor of 3.3 to 3.6W to fail. However, because of the location of the container in the forward hold, the rough seas experienced during the storm and the speed of the ship, the container would have been subjected to multiple high impact loads because of the ship's pitching and pounding. Further, given that containers are free to move within their guides, and that only one container was stacked on top of the failed container, it is possible that the failed container was able to move vertically from its footings while the vessel was pitching and pounding, causing significant vertical impact loads. The failed container was filled to 77percent of its capacity. As per the IM101 requirements, tanks containing between 20 and 80percent of their capacity should not be offered for transport by ship, and, therefore, the failed container was marginally underfilled. This could have resulted in liquid surge, shifting the centre of gravity to one end. This effect, combined with vertical acceleration, can significantly increase the vertical loads at one end. The IMDG Code has a similar requirement, although the exact ullage allowance must be determined using a formula which takes into account the temperatures and expansion characteristics of the contents. The damage pattern on the container was consistent with the damage having occurred as a result of operational loads during a storm at sea. However, the possibility exists that the longitudinal I-beams may have been slightly buckled or otherwise damaged prior to loading. If a buckle was present on the longitudinal I-beam prior to loading aboard the ship, the load required to cause the type of failure observed would have been much less than on an undamaged frame. To reduce the damage resulting from handling in the railway transportation system, and to facilitate observation of such damage, the container manufacturer has instituted a 36-ton upgrade of the unit by welding diagonal braces in two additional holes in the web, stiffening the latter and increasing its buckling strength. To date, more than 46percent of the 1,448 tank containers produced in the 30-ton series have been upgraded to the 36-ton level. At this time, there has been no reported failure of the new 36,000kg-rated tank container frame. The chemical that spilled into the No.1 hold of the MORU.K. is listed in the IMDG Code published by the International Maritime Organization (IMO). Its properties are given under numerical index UNNo.1760. Its safety data sheet indicates that it is a Class8 product, which means it is corrosive. This solution is composed of sodium borohydride, sodium hydroxide and water, and its properties are to react chemically when exposed to metals to release hydrogen. The vessel sailed westward head on to the wind at an estimated speed of 10 knots. A spill of a corrosive chemical occurred in No.1 hold when the ship encountered adverse weather conditions. The vessel had to discharge an estimated eight tonnes of the chemical into the sea, with the permission of the Canadian Coast Guard, in order to stop the formation of hydrogen in the holds, as it might have caused an explosion. If the tank had been loaded to its maximum permissible weight of 29,030kg, excluding the weight of the frame, a vertical inertial load factor of between 3.25 and 3.5W would have been required to cause this failure. The metal support structure of the tank container was damaged by the acceleration forces generated by the heaving and pounding of the vessel in the storm. Established norms require that a container be filled to 80percent of its capacity for shipment by sea. At 77percent of its capacity, the failed container was marginally underfilled.Findings The vessel sailed westward head on to the wind at an estimated speed of 10 knots. A spill of a corrosive chemical occurred in No.1 hold when the ship encountered adverse weather conditions. The vessel had to discharge an estimated eight tonnes of the chemical into the sea, with the permission of the Canadian Coast Guard, in order to stop the formation of hydrogen in the holds, as it might have caused an explosion. If the tank had been loaded to its maximum permissible weight of 29,030kg, excluding the weight of the frame, a vertical inertial load factor of between 3.25 and 3.5W would have been required to cause this failure. The metal support structure of the tank container was damaged by the acceleration forces generated by the heaving and pounding of the vessel in the storm. Established norms require that a container be filled to 80percent of its capacity for shipment by sea. At 77percent of its capacity, the failed container was marginally underfilled. The failure of the metal support structure of the tank container was probably caused by the combination of several factors, including the storage location of the containers in the forward hold, the severe storm conditions, the speed of the ship during the storm, the marginal ullage conditions, and possible pre-loading damage.Causes and Contributing Factors The failure of the metal support structure of the tank container was probably caused by the combination of several factors, including the storage location of the containers in the forward hold, the severe storm conditions, the speed of the ship during the storm, the marginal ullage conditions, and possible pre-loading damage.